In February we've announced our partnership with SafeDX data center co-founded together by Foxconn and Intel to provide Signals with a powerful infrastructure to collect and process terabytes of data from cryptotrading markets. These days, SafeDX is implementing a new Intel Rack Scale Design (Intel RSD) solution, first of its kind in Europe. We are proud to be the first company using their own OpenStack based on this revolutionary infrastructure and present our unique use case on Intel Partner Connect Europe in May.
Ideal infrastructure for advanced trading algorithms
Intel RSD architecture enables a dynamic composition of resources to meet specific workload requirements. This is the ideal for demanding computations using algorithms such as LSTM Neural Network or recurrent neural networks in their input and predict the next value. Signals are working in collaboration with SafeDx on using this scalable infrastructure for prediction of trading time series from crypto markets in real time.
Signals philosophy is to encourage traders, developers, and data scientists to create new trading models by providing them with the best tools on the market. Signals platform shields traders without coding skills from the implementation details signals Strategy Builder. In the same way, Signals shields data scientists and developers from the underlying infrastructure and enables them to fully focus on trading models and algorithms. Providing this new infrastructure to your trading algorithms without wondering about scalability or performance.
Together with our partner Quote from Intel Partner Europe conference in Prague, May 15-17, 2018. Stay tuned, we will keep you updated!
Working hand in hand, Signals and SafeDX data center can efficiently collect and pre-process big data from various crypto-markets in real time.
To enable the storage and processing of terabytes of data within the cryptotrading market, as well as media analyzes and blockchain monitoring, Signals is collaborating with SafeDX.
SafeDX is a powerful data center with related computational infrastructure based in Prague, the capital of Czech Republic. Co-founded by Foxconn, SafeDX is part of the Foxconn group's international network that offers data center services. It specializes in providing big data IT services with an excellent level of security and flexibility, as well as low operating costs. This makes SafeDX the perfect match for Signals' goal to run complex data operations in a short amount of time. The goal of this partnership is to enable Signals to download and process terabytes of data every day. By data, we refer to the data within the crypto trading market - data related to sentiment analysis and blockchain inspection.
"Vladislav Kral, project director at Foxconn Technology CZ, said:" Vladislav Kral, project director at Foxconn Technology CZ, says the computing power that we have put in place based on our strategic relationship and collaboration with Intel. "
SafeDX primarily focuses on the Central and Eastern Europe region, but it is perfectly capable of supporting the activities of clients anywhere in Europe and Asia. Signals' partnership with Foxconn, together with the cooperation with Intel, will create the infrastructure necessary for a platform of data storing, scaling and speed of processing. This partnership is crucial for collecting and pre-processing large amounts of data from various crypto-markets in real time. It enables cryptotraders to make smarter, faster trading decisions and to maximize their profits.
Foxconn does not need any introduction in the crypto community. Besides being the largest manufacturer of electronic products and providing worldwide data center services, they offer out of the box hardware configurations for crypto mining. Another noteworthy fact: Foxconn invested in blockchain technology and even its own subsidiary blockchain company, called Chained Finance.
We 'll like to fill you in on how the SGN discount structure works during the upcoming Signals Token Sale.
A visualization of the SGN token discount structure.
SGN token discount structure
The price of the 1 SGN token without any discount is set at 0.00036 ETH ($ 0.36 with the Ether price fixed at $ 1000). In the main Token Sale, there is a discount structure based on the amount of tokens that have already been sold. The discount structure functions in the following way: The first participant of the Token Sale will have a discount of 15%. This price will be USD. USD raised during the sale.
Once we've reached the middle of the Token Sale (participate in the sale is worth $ 9M), participants will receive a discount of 7.5%. It is also important to mention that the first participant buys SGNs for 100 ETH, the discount of 15% is applied to the total amount in this purchase. Thus, it is beneficial for the participants to make a single discount purchase and receive a less discount.
The SGN Token Sale starts on March 12, 2018
Make sure to register and go through our Know Your Customer (KYC) process before the Token Sale begins - this way, you can avoid the crowds at the start of the sale. You will need to go through the KYC process even if you are already registered during autumn 2017 and participated in the SGN token presale.
The final countdown to the SGN Token Sale starts now! Stay tuned for more news leading up to the sale from our team.
If you go to college and take a course "Machine learning 101", this might be the first example of your learning machine teacher will show you:
Imagine you work for a real estate agency, and you want to predict, for how much a house will sell. You have some historical data - you know that house A has been sold for $ 500 000, house B for $ 600 000, and house C for $ 550 000. You know what about the house in square meters, number of rooms in the house, and the year the house was build.
The goal of the real estate agency is to predict, for how much a new house D will sell, given its known properties (size, age and number of rooms of the house). In ML terminology, the known properties of the house are called "features" or "indicators" (we use the term "indicators" in Signals, since this term has been historically used in trading). The price of the house is your "target". [3] [9]
Let's look at the data with a human eye:
Each row is a training example, and it contains three indicators and one target value. You might find that smaller houses are cheaper, and that newer houses are more expensive. You can use this data gained from historical data and if a new house comes to your agency, you might price it accordingly.
OK, but what if you have much more data, let's say hundreds of thousands of houses? You as a human will never be able to process such data. Another problem can appear if you have lots of indicators - not only number of rooms, size and age of the house, but let's say thousands of indicators. Your human mind will have great difficulties to reveal relationships between these features. It may take you months of trying to understand the data, and still you might just project your false assumption - for example, you believe that bigger houses are more expensive before you saw the data, and you will be tempted to believe it even if the data say otherwise. [5]
What if someone else can learn from this data, someone who is much better suited for processing huge and somewhat boring structured data? An algorithm? A "machine"?
It turns out that there are such algorithms. Machine learning algorithms, which accept the data in the format we have shown above, learn from these data (and we can say that we "train" the algorithm on the data, so these data are called "training data"), and when they later receive a new, unseen example, they output a prediction. These algorithms can be as simple as linear regression and as complex as neural networks, but it is just mathematics. The main idea behind all these algorithms is optimizing on the known data to find a function (linear in the case of linear regression or quite complex in the case of neural networks), which fits the data well but not too much to "overfit". This fitted function is then used to predict the "target" for the unseen data. [6] [12] [8]
How do we use machine learning in Signals?
Machine learning is much more than the simple example described above. In Signals, we use ML in the following ways:
1. Strategy optimization
Even if you decide not to use machine learning and to define your strategy manually, methods from computer science and statistics, which are closely related to machine learning, can help you.
In your strategy, each indicator has several parameters. You might use a random set of parameters, or you can try to grid-search through all the parameters and use parameters which perform best on historical data. The problem is, the first approach never works, and the latter approach becomes computationally unfeasible if you have more than just a few parameters.
This problem is called optimization and is well-studied. In Signals, we implement genetic algorithms for parameter optimization, and in the future we plan to implement other methods, such as bayesian optimization. [4] [12]
2. Signals extraction
This use of ML is most similar to the article. In Signals extraction, the data we use are time-series data, such as bitcoin price chart. The user selects indicators (features) from Indicator Marketplace and feeds their extracted from time series to machine learning algorithm. The ML algorithm then learns from the data, finds non-linear relationships between the indicators, and predicts the target value on the data. [10] [16]
How does Signals extraction works?
Time series preprocessing is needed in many other fields than in trading - speech, audio or accelerometer signal processing, weather forecasting, ...
In trading, technical analysis indicators are popular. There are many reasons to learn from the traders. You can find them implemented in most trading software.
Traders mostly use these indicators to indicate. [7] [2] At Signals, we provide algorithms for algorithm.
We call this feature Signals Extraction, users select the combination of theories which they want to use in their model. [9] [13] Once the signals are performed on historical data (let's say January to May), the Signals platform evaluates its performance on unseen historical data. and profit.
This ML playground will enable users to experiment with multiple ML algorithms with different subsets of indicators and use only the algorithm which performed well on unseen historical data to make money in the real world!
Signals give you the playground and the experts take care of the data flow, so you will not make some stupid mistake - as implementing the features / algorithm incorrectly and give them completely unmeaningful parameters, or as predicting the past.
However, there are many things you will have to decide for yourself.
Which combination of technical indicators will you try, and what kind of parameters for the indicators will you use (e.g. sizes of time window)?
Will you try some feature (indicator) selection / transformation algorithm?
What kind of machine learning algorithm will you try? What parameters of these algorithms?
Will you learn on the whole historical data, or just on past few months? Or will you always learn only on the past few days and predict the next day?
What will be your target? How will you define the buy / sell signal?
There are so many options and so much data, that you will not be able to try all of them. That's what makes algorithmic trading so addictive.
3. Indicators based on Machine learning
In the example of toys with home prediction, there is one important thing to note: the indicators are designed by humans. One must decide that the algorithm will learn to predict the price of the size of the house, the number of rooms and the age of the house, and not from other known properties, such as the first letter of the street or the second letter of the city. The technical indicators described above are also designed for humans, and although they work very well in many ways, they may work better with other features - features based on machine learning.
Machine learning techniques can be used for
1. study the full indicator of unstructured data, use for example a single layer or deep neural network. [1]
2. create indicators using advanced ML techniques, such as when you use the Natural Language Processing method for media sentiment analysis and social networking.
Your machine learning Predictor can use this ML indicator in addition to technical indicators. One of your new indicators, which you will feed the ML predictor, may be the sentiment (mood) of bitcoin on Twitter in the last 10 minutes (on a scale of 1-10) and other new indicators may be representations of neural networks (= vector n- dimmensional) learning from the Ethereum timeline in the last 15 minutes.
Of course, you can decide to fully trust the machine and use only machine-based indicators to feed your ML predictors! On the other hand, you can add some human engineering features to the indicator-learning algorithm - for example, you might feed them with data changed by Fourier transforms instead of raw signals.
There are also machine learning algorithms designed to work with raw time series - they take the time series as input and predict the next value. Probably the most popular of these algorithms is LSTM. LSTMs are difficult and expensive to train (which can actually be your advantage in the trading market!), And they work really well for some time series issues. You can experiment with LSTMs in Signals. [11] [14] [15]
One of the best things about Signals is that you can apply your own indicators. 20% token will be used to support the Data Sciences community and we are excited to see what new indicators the community will bring to the platform as we have received many collaborative requests from developers of the data science community.
Of course, each indicator must apply the Signal indicator specification so that it can be automatically used as a graphical component in the Signal Strategy Maker and used by cryptotraders.
[1] A. Coates, H. Lee, and A. Ng., "Single-Layer Network Analysis in Unattended Learning Learning", JMLR Workshop and Conference Proceedings, vol. 15, p. 215-223., 2011.
[2] A. N. Azizan and J. C. P. Mng, "Can technical analysis predict price futures?", IUP Journal of Financial Risk Management, vol. 7, no. 3, p. 57-75, Sep. 2010
[3] A. W. Lo and A. C. MacKinlay, "Stock market prices do not follow a random path: Evidence from a simple specification test," Rev. Financial Stud., Vol. 1, no. 1, p. 41-66, Jan. 1988.
[4] D. Barber, Bayesian reasoning, and machine learning. Glasgow, U.K.: Cambridge University Press, 2012.
[5] G. Friesen and P. A. Weller, "Quantify the cognitive bias in the analyst's earnings forecast," J. Financial Mark., Vol. 9, no. 4, p. 333-365, November 2006.
[6] I. Kaastra and M. Boyd, "Designing a neural network to forecast a series of financial and economic times," Neurocomputing, vol. 10, no. 3, p. 215-236, April 1996.
[7] J. Stanković, I. Marković and M. Stojanović, "Optimization of Investment Strategies Using Technical Analysis and Predictive Modeling in Emerging Markets" Procedia of Economics and Finance, vol. 19, p. 51-62, 2015.
[8] K. P. Murphy, Machine learning: A probabilistic perspective. Cambridge, MA, USA: MIT Press, 2012.
[9] M. T. Leung, H. Daouk, and A. S. Chen, "Forecasting stock indices: Comparison of classification models and level estimates," Int. J. Forecasts., Vol. 16, no. 2, pp 173-190, April-June. 2000.
[10] NI Indera, IM Yassin, A. Zabidi, and ZI Rizman, "Non-linear autoregressive with an exogenous bitcoin bitcoin price prediction (narx) model using optimized PSO parameters and moving average technical indicators," J. Fundam . Appl. Sci., Vol. 9, no. 3 S, p. 791-808, September 2017.
[11] Ordóñez FJ, Roggen D. Deep Convolutional and LSTM Recurrent Neural Networks for Multimodal Wearable Activity Recognition. Liu Y, Xiao W, Chao H-C, Chu P, eds. Sensors (Basel, Switzerland). 2016; 16 (1): 115. doi: 10.3390 / s16010115.
[12] S. A. Mitilineos and P. G. Articists, "Forecasting future stock prices using artificial neural networks and genetic algorithms," Int. J. of Decision Sciences, vol. 7, no. 1/2, pp. 2-25, April 2017.
[13] S. Thawornwong and D. Enke, "Selection of adaptive financial and economic variables for use with artificial neural networks," Neurocomputing, vol. 56, p. 205-232, January 2004.
[14] T. Fischer and C. Krauss, "In-depth learning with short-term memory networks for financial market predictions," Eur. J. Oper. Res., 1-16, January 2018.
[14] X. Pang, Y. Zhou, P. Wang, W. Lin, and V. Chang, "Innovative neural network approach to stock market prediction," J. Supercomput., Pp. 1-21, January 2018.
[16] Y. Shynkevich, T. M. McGinnity, S. A. Coleman, A. Belatreche, and Y. Li, "Forecasting price movements using technical indicators: Investigate the impact of various lengths of the input window," Neurocomputing, vol. 264, p. 71-88, November 2017.
WEBSITE WHITEPAPER
TWITTER FACEBOOK
TELEGRAM
Tidak ada komentar:
Posting Komentar